Note

1. Expore CH$_{4}$ Laboratory Spectra¶

$xpectra_flowchart_1.jpg$

Table of Contents¶

1.1 | Load the laboratory spectra
1.2 | Check if there is any negative or NAN value
1.3 | Visualize the spectra
1.4 | Save it to the output directory

1.0 - Ensure Xpectra has been correctly installed¶

Created the Xpectra_project directory
Downloaded reference_data from this link
Put the datasets, figures, and tutorials/notebooks are in your Xpectra_project/reference_data directory.
Defined the path to your reference_data. Confirm it using os.getenv("Xpectra_reference_data"), but if you encounter an error, simply hard code it by defining the path as __reference_data_path__.
Cloned Xpectra using git clone https://github.com/EhsanGharibNezhad/Xpectra.git
Lastly, check you are able to execute import Xpectra

[1]:

# Import necessary modules
from Xpectra.SpecFitAnalyzer import *
from Xpectra.SpecStatVisualizer import print_spectral_info, plot_spectra_errorbar_seaborn

1.1 - Import CH4 lab spectrum¶

$\rightarrow$ After this setup, we can now import the $\mathrm{CH_{4}}$ spectrum as a DataFrame and seperate into two numpy arrays: x (wavenumber [$cm^{-1}$]) and y (signal).

[2]:

# Call environment variable and assign path to data
__reference_data_path__ = os.getenv("Xpectra_reference_data")
print(__reference_data_path__)

/Users/egharibn/RESEARCH/ml/projects/Xpectra_project/reference_data

[3]:

methane_spectrum = pd.read_csv(os.path.join(__reference_data_path__, 'datasets','Spectrum_CH4_100Torr.csv'))

# Create x and y arrays
x = 10**7/methane_spectrum['W'].to_numpy() # Wavenumber
y = methane_spectrum['I'].to_numpy() # Intensity

1.2 - Check if there is any negative or NAN value in the data¶

Xpectra.SpecFitAnalyzer module has 3 essential purposes: process spectral data (current), fit and correct spectral baseline, and identify and fit spectral peaks.

[4]:

# Initialize SpecFitAnalyzer
specfit = SpecFitAnalyzer(wavenumber_values = x, signal_values = y)

$\rightarrow$ Scan arrays for negative or NAN values, print results, and save to class

[5]:

# Check for NAN or negative values
specfit.check_negative_nan()

No NAN values.
9614 negative values found (7.78% of data)

$\rightarrow$ Values now saved as instance attributes:

[6]:

# Print attributes
print("x_cleaned:",specfit.x_cleaned)
print("y_cleaned:",specfit.y_cleaned)
print("negative_indeces:",specfit.negative_indeces)
print("nan_indeces:",specfit.nan_indeces)

x_cleaned: [2898.54306009 2898.54390792 2898.5447659  ... 2985.05887607 2985.05951532
 2985.0600492 ]
y_cleaned: [0.44029694 0.4364364  0.43730026 ... 0.45600653 0.45858333 0.45689613]
negative_indeces: [   193    194    195 ... 120542 120543 120544]
nan_indeces: []

[7]:

# Convert spectra to absorption
y = np.exp(-y)

1.3 - Visualize the spectrum¶

$\rightarrow$ Print spectral information

[8]:

print_spectral_info(x, y, print_title='Methane Lab Spectrum')

******************************

Methane Lab Spectrum

{   'Number of points': 123527,
    'Wavenumber range (cm-1)': (   2898.5430600916807,
                                   2985.060049195717)}
******************************

$\rightarrow$ Plot spectrum using Seaborn, and save

[9]:

plot_spectra_errorbar_seaborn(wavenumber_values = x,
                              signal_values = y,
                              __reference_data__ = __reference_data_path__,
                              absorber_name = symbol_dict['CH4'],
                              plot_type = 'line',
                              __save_plots__ = True)

../_images/tutorials_1_Explore_Lab_Spectra_21_0.png

$\rightarrow$ Plot spectrum interactively using Bokeh, enabling zoom and panning

[10]:

plot_spectra_errorbar_bokeh(wavenumber_values = x,
                            signal_values = y,
                            absorber_name = 'CH4',
                            plot_type = 'line')

Loading BokehJS ...

1.4 - Save the processed spectra¶

$\rightarrow$ Create DataFrame with original and processed spectra

[11]:

# Create DataFrame
data = {
    'original_x': x,
    'original_y': y,
    'cleaned_x': specfit.x_cleaned,
    'cleaned_y': specfit.y_cleaned
}

df = pd.DataFrame.from_dict(data, orient='index').transpose()

[12]:

display(df)

	original_x	original_y	cleaned_x	cleaned_y
0	2898.543060	0.643845	2898.543060	0.440297
1	2898.543908	0.646336	2898.543908	0.436436
2	2898.544766	0.645778	2898.544766	0.437300
3	2898.545133	0.639101	2898.545133	0.447693
4	2898.545638	0.630384	2898.545638	0.461426
...	...	...	...	...
123522	2985.057227	0.635260	NaN	NaN
123523	2985.058132	0.635798	NaN	NaN
123524	2985.058876	0.633810	NaN	NaN
123525	2985.059515	0.632179	NaN	NaN
123526	2985.060049	0.633246	NaN	NaN

123527 rows × 4 columns

$\rightarrow$ Save the spectra to CSV file

[13]:

df.to_csv(os.path.join(__reference_data_path__,'processed_data','processed_methane_spectrum.csv'), index=False)

1. Expore CH\(_{4}\) Laboratory Spectra¶